US20060288126A1

US20060288126A1 - Using a processor to program a semiconductor memory

Info

Publication number: US20060288126A1
Application number: US11/499,340
Authority: US
Inventors: Peter Mueller
Original assignee: Marvell International Ltd
Current assignee: Marvell Asia Pte Ltd
Priority date: 2003-12-23
Filing date: 2006-08-04
Publication date: 2006-12-21
Anticipated expiration: 2023-12-23
Also published as: US20050138301A1; US7325087B2; US7594065B2; US20060259653A1; US7099985B2

Abstract

A memory programmer may be coupled through a first processor and a physical interface to a semiconductor memory to be programmed. The interface may be the same interface that allows two separate processors in a multiprocessor memory to communicate with one another in one embodiment. Thus, an independent memory bus coupled directly to the memory components to be programmed may be eliminated, reducing form factor, decreasing costs, and increasing manufacturing throughput in some embodiments of the present invention.

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application is a continuation of U.S. patent application Ser. No. 10/744,035, filed on Dec. 23, 2003.

BACKGROUND

This invention relates generally to programming semiconductor memories.
Many microprocessors are utilized in embedded applications, such as automobile systems, stereo systems, watches, personal digital assistants, a variety of smart industrial machines, and smart telephones, to mention a few examples. In many of these embedded applications, microprocessors need to exchange large amounts of data with other microprocessors, system level peripherals, and external servers. In addition, a large amount of data of different types may be exchanged from processors, including both command and control information, databases, streaming data, and large file transfers.
A high speed serial link may be utilized to enable communications between embedded processors. One such high speed link is called the Mobile Scalable Link (MSL) available from Intel Corporation, Santa Clara, Calif. The MSL may reduce the bottleneck of data exchanged between general purpose application processors and baseband processors in next-generation hand held mobile devices such as smart phones and personal digital assistants. A number of semiconductor memories may be coupled to application processors.
Generally those memories are programmed through a memory bus coupled directly to the packaging for the integrated semiconductor memory. The memory bus includes a large number of pins, increasing the size of the integrated circuit package for the memory and the cost of the overall system.
Thus, there is a need for alternate ways to program semiconductor memories in processor-based integrated circuit systems.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a system depiction of one embodiment of the present invention;
FIG. 2 is a schematic depiction of one embodiment of the serial link shown in FIG. 1 in accordance with one embodiment of the present invention;
FIG. 3 is a software stack for one embodiment of the serial link shown in FIG. 1;
FIG. 4 is a flow chart for software for one embodiment of the present invention; and
FIG. 5 is a partial top plan view of another embodiment of the present invention.

DETAILED DESCRIPTION

Referring to FIG. 1, a system 10 may be a mobile embedded system in one embodiment of the present invention. The system 10 may include an application or general purpose processor 22 coupled through a physical interface 24 to a board or socket 21. The processor 22 may be an embedded processor in one embodiment.
In some embodiments, a second processor, not shown in FIG. 1, may be coupled to that board or socket 21. In one implementation of the present invention, the application processor 22 may operate through the interface 24 to exchange data with an onboard or in-socket baseband processor not shown in FIG. 1. Applications for such systems include smart telephones, cell phones, and personal digital assistants, to mention a few examples.
The application processor 22 may be an integrated circuit with internal memory 20 in one embodiment of the present invention. The application processor 22 and internal memory 20 may communicate with an integrated memory component 18. Thus, a stacked component or multi-component package (MCP) 12 may include the packaged integrated circuits 26 and 28 in one embodiment of the present invention.
The integrated memory components 18 may be stacked flash memory integrated circuits in one embodiment of the present invention. However, other non-volatile or volatile semiconductor memories may also be used, including electrically eraseable read only memory, polymer memory, ovonic, or phase change memory, and static random access memory, to mention a few examples.
The integrated circuits 26 and 28 may be separately packaged in one embodiment. In another embodiment, the circuits 26 and 28 are contained within the same package. In still other embodiments, the components 18 and 22 may be part of the same integrated circuit. In some cases, the board or socket 21 may not be used.
The integrated memory components 18 would have been programmed conventionally by connecting a memory bus (not shown) from the components 18 to a memory programmer 16 that may be part of a manufacturing system 14. However, such a memory bus commonly uses around 50 pins. This increases the form factor of the package for the components 18 and thereby increases cost.
However, in the system 10 shown in FIG. 1, the memory programmer 16, which may be part of a manufacturing system 14, may be plugged into the board or socket 21 to communicate with the integrated memory components 18 through the application processor 22 and its physical interface 24. Thus, the memory programmer 16 may program the components 18 indirectly through the application processor 22. Effectively, in one embodiment, the application processor 22 may handle high speed data from the interface 24, buffer that data, and then feed the buffered data, at a slower rate, to the memory to be programmed.
In some embodiments of the present invention, software 48 may be executed by the application processor 22 to facilitate the programming of the components 18. As a result, it is no longer necessary to couple a memory bus to the components 18. Instead, a memory bus connection 29 may be established between the processor 22 and the component 18. In some embodiments, this may reduce form factor, decrease cost, and increase programming speed, which reduces manufacturing program time. Increasing programming speed may increase manufacturing throughput and further decrease costs in some embodiments.
Referring next to FIG. 2, the physical interface 24 may be a high speed, high bandwidth serial link that, in one embodiment, may be the Mobile Scalable Link from Intel Corporation. As another example, the interface 24 may be one compliant with the Universal Serial Bus Revision 2.0 specification (Apr. 27, 2000) available from the USB Implementers Forum, Inc., Portland, Oreg. 97221. The interface 24 may couple the application processor 22 with a baseband processor (not shown). For example, in one embodiment, the application processor 22 may include a processor internal connection 22 a and the board 21 may have a similar connection 22 b.
The interface 24, in one embodiment, may include a pair of unidirectional, high speed links 36 and 38 for connecting two nodes made up of connections 22 a and 22 b.
Transmit first-in-first-out (FIFO) buffers 30 may couple to an outbound link 36 and to the internal connection 22 a in one embodiment. An inbound link 38 may couple to the receive first-in-first-out (FIFO) buffers 32 to the internal connection 22 a. Each link 36 and 38 may receive a clock signal, one to four data channels, a channel identifier, and a wait pin, as indicated by FIG. 2. While a hardware depiction for the interface 24 is illustrated, the present invention is in no way limited to a particular hardware architecture.
Referring to FIG. 3, a software stack for the interface 24 may be logically divided into three layers 42, 44, and 46 in one embodiment. The logical link control layer 42 may provide general framing to higher layer packets and may help establish connections between nodes. The physical link control layer 44 may multiplex higher layer packets over the available interface physical channels and provide quality of service functionality. The physical link manager layer 46 may interface with low level platform elements, like direct memory access and interrupts, and may help in actual data transfer. The physical link manager layer 46 may also deal with overall link configuration and power management. While a software stack is depicted in FIG. 3, the present invention is in no way limited to any particular software architecture.
In some embodiments, a memory bus need not be connected to external pins on the package for the integrated memory components 18. Instead, the external manufacturing system 14 may provide programming data over the physical interface 24 and the application processor 22 then programs the memory components 18.
In one embodiment, the components 18 may be effectively stacked on one another and on top of an integrated circuit 26, including the application processor 22. This arrangement may result in more efficient communications between the various components. The components 18, the memory 20, and the processor 22, when implemented on different integrated circuits, may be packaged together in one integrated circuit package. However, the present invention is not limited to any type of stacking or packaging of the various elements.
Software 48 for facilitating the programming of the integrated memory components 18 may be stored on the application processor 22 in one embodiment. However, the software 48 may be located in the internal memory 20 or at any other location. In one embodiment, the software 48 is stored in non-volatile memory.
Referring to FIG. 4, the program software 48 begins by receiving high speed transfer of programming data from the memory programmer 16 over the interface 24 as indicated in block 50. The software 48 causes the application processor 22 to buffer the received data in a volatile memory as indicated in block 52. The volatile memory may be part of the internal memory 20 in one embodiment of the present invention. The buffer data is thereafter provided to the memory components 18 to be programmed as indicated in block 54. The provision of such data may be at a rate commensurate with the rate of programming of the integrated memory components 18. That rate may be significantly slower than the rate at which the data may be transferred from the memory programmer 16 across the interface 24 and processed by the processor 22 in some embodiments. Thus, in such embodiments, the processor 22 may transform the speed of data transferred to accommodate the integrated memory components 18, while at the same time allowing the memory programmer 16 to transfer the data quickly. Transferring the data quickly increases the manufacturing throughput.
Referring to FIG. 5, in accordance with one embodiment of the present invention, the board or socket 21 may have received thereon a large number of multi-component packages 12, each including a processor 48 and integrated memory components 18. The packages 12 may receive power from the board or socket 21. Each processor 22 may be coupled by an appropriate electrical connection to the memory programmer 16.
In one embodiment of the present invention, the memory programmer 16 programs one or more of the packages 12 at a time. Data may be streamed to some number less than all of the packages 12 and then, once the programming data has been transferred, the memory programmer 16 may be free to move on to program other packages 12 on the board or socket 21.
Thus, in one embodiment of the present invention, the programming data may be transferred at a high speed from the programmer 16 for processing by the application processor 22. Thereafter, the memory programmer 16 may subsequently provide data to a different package 12. While the memory programmer 16 is continuing to access subsequent packages 12 and potentially thereafter, the processor 22 functions to transfer the buffered data to actually program the integrated memory components 18.
In other words, the programming of the integrated memory components 18 may proceed during and after the actual provision of data from the memory programmer 16. This may free the programmer 16 to move on to other packages 12 and, subsequently, to other boards or sockets 21 if need be. As a result, manufacturing throughput may be increased in some embodiments.
While the present invention has been described with respect to a limited number of embodiments, those skilled in the art will appreciate numerous modifications and variations therefrom. It is intended that the appended claims cover all such modifications and variations as fall within the true spirit and scope of this present invention.

Claims

1. A method comprising:

programming a semiconductor memory by providing the programming data to a processor to provide the data to the semiconductor memory.

2. The method of claim 1 including transferring data from a memory programmer across a physical interface to an integrated circuit including a processor.

3. The method of claim 2 including buffering said programming data on said integrated circuit including said processor.

4. The method of claim 2 including enabling said processor to buffer said programming data.

5. The method of claim 4 including transferring the data to a buffer storage and subsequently transferring the data from said buffer storage to said memory to be programmed.

6. The method of claim 5 including transferring said data from said buffer storage to said memory to be programmed at a slower rate than said data is transferred by said memory programmer to said integrated circuit.

7. The method of claim 5 including transferring the data from a first integrated circuit including said processor to a second integrated circuit including said semiconductor memory to be programmed.

8. The method of claim 5 including transferring the data from the memory programmer to said processor using a board to receive a large number of memories to be programmed, said memories being coupled to processors to receive the data to be programmed.

9. The method of claim 8 including programming the memory after the data has been transferred from said memory programmer and buffered.

10. The method of claim 9 including successively transferring data to a number of memories to be programmed, maintaining power to said memories and programming said memories using said processors after said data has been buffered.

11. A semiconductor memory comprising:

a permanent connection to another integrated circuit including a processor, said connection including a memory bus to couple said processor to said memory.

12. The memory of claim 11 wherein said memory is a flash memory.

13. The memory of claim 11 wherein said memory includes a stack of integrated circuit chips.

14. The memory of claim 13 wherein said chips are packaged in one package including said connection.

15. The memory of claim 14 wherein said package includes no pins to receive a memory bus.

16. A system comprising:

a processor to buffer memory programming data received from a serial link; and

a memory to be programmed by the processor using the programming data.

17. The system of claim 16 wherein said memory is a flash memory.

18. The system of claim 16 wherein said memory includes a stack of integrated circuit chips.

19. The system of claim 16 wherein said memory is packaged separately from said processor and said system including a permanent connection to connect said memory and processor packages.

20. The system of claim 16 wherein said serial link to transfer data at a rate higher than the rate at which said memory may be programmed.